Snakemake

A workflow tool similar to GNU Make, but with much more power with Python.

• https://github.com/snakemake/snakemake

Examples

• PyCon Korea 2015: 탐색적으로 큰 데이터 분석하기

Tips

General

Rerun upon input changes

Snakemake doesn’t automatically rerun rules when input files are added. But the rerun can be triggered by snakemake -n --forcerun $(snakemake --list-input-changes)

There are also --list-code-changes, --lc, --list-params-changes, --lp.

script directive

when used, it injects snakemake variable. For instance, input files can be accessed via snakemake.input[0]. It is convenient, but it makes the script not technically correct and may make it harder to test.

Temporary & protected output files

You can use temp() or protected() in output directive.

See https://snakemake.readthedocs.io/en/stable/tutorial/advanced.html#step-6-temporary-and-protected-files

Print command

-p option prints the shell commands. It can be useful for debugging.

Modularization

You can either include another snakefile or use another (complete) Snakefile as a sub-workflow.

Parameterization

See Parameter space exploration

Formatting

snakefmt

There is a package called snakefmt that is derived from Black.

Flake8

To remove the errors/warnings from flake8 about the variable snakemake that snakemake injects into scripts, you can add to builtins = snakemake to ~/.config/flake8 or run with the option --builtins=snakemake.